Nvidia Cuts Vera Rubin HBM4 Bandwidth 13% [2026]
Nvidia reduced memory and interconnect bandwidth targets for its Vera Rubin AI hardware, with HBM4 bandwidth dropping 13% to 19.2 TB/s and NVLink 6 bandwidth falling 16.7% to 3 TB/s. The changes affect all configurations, including the 72-GPU rack. Nvidia stated the original figures were preliminary targets, and the new specs reflect production capabilities.
How this was made
![Nvidia Cuts Vera Rubin HBM4 Bandwidth 13% [2026] — source image](/_next/image?url=%2Fdata%2Fnews-image%3Furl%3Dhttps%253A%252F%252Fshattered.io%252Fwp-content%252Fuploads%252F2026%252F11%252Fnvidia-vera-rubin-bandwidth-cut-hbm4-19-2-tbs-nvlink-2026-gen.webp&w=2048&q=75)
The 30-second read
Why it matters
Reduced bandwidth targets could affect revenue forecasts for Nvidia's AI segment and alter competitive dynamics with AMD.
Market read
First‑time disclosure of spec reductions for a flagship AI product, potentially influencing Nvidia's stock and AI‑hardware sector sentiment.
What to watch
Supply constraints at SK Hynix and Samsung may have forced the change, indicating broader memory‑chip market pressure.
Background
Nvidia's Vera Rubin AI platform is a key component of its data‑center strategy, with customers like Samsung SDS planning GPU‑as‑a‑service launches.
Ticker impact
Nvidia announced reduced HBM4 bandwidth targets for its Vera Rubin AI hardware, cutting per‑GPU bandwidth by ~13% and rack bandwidth by up to 26%.
likely downward pressure as the market prices in weaker performance specs
Spec reductions affect perceived competitive advantage versus AMD and could trigger short‑term sell‑offs.
Market effects
AI hardware sector may see re‑pricing of competing vendors as Nvidia's specs are revised.
US tech equities could face modest pullback, especially AI‑focused ETFs.
Limited to investors tracking Nvidia and AI hardware supply chain.
Counterpoint
The bandwidth cut may be minor if real‑world performance remains strong; investors could view it as a realistic adjustment.
Key entities
- CompanyNvidia
US‑listed semiconductor and AI hardware maker
- CompanySamsung SDS
Partner planning Vera Rubin GPU‑as‑a‑service launch
